USDOT could advance travel modeling and help planners account for induced demand

By Chris McCahill 

A provision of the Bipartisan Infrastructure Law (Sec. 11205) requires USDOT to review existing travel demand models and, among other things, consider the potential implications of induced travel. Federal officials, committed to that mandate, were at the TRB annual meeting last week to learn from modeling experts and practitioners. This blog post offers one perspective on the issues and lays out several opportunities gleaned through discussions at TRB. 

Induced travel is a well-documented phenomenon whereby new highway capacity causes traffic to increase by around the same proportion. A good rule of thumb is that a 10% increase in highway capacity will lead to a 5% traffic increase within a few years and up to 10% within 10 years. Plenty of studies from across the U.S. back this up, as we’ve written before. 

An important question for USDOT and other transportation professionals: Do traffic forecasts properly account for the effects of induced travel? Some advocates argue that agencies ignore induced demand or that it is too hard for policymakers to understand. But, as research looking at traffic forecast accuracy suggests, that is not entirely true. There is a strong possibility that induced travel is baked into every decision-making process, to such an extent that many of us take it for granted. 

A comprehensive NCHRP study suggests most agencies overestimate traffic growth. The researchers behind that study compared traffic forecasts from nearly 1,300 projects to actual traffic volumes, years later. They found there is around 6% less traffic than predicted, on average. Similarly, national traffic forecasts have vastly overestimated traffic growth for decades (shown below). In other words, transportation planners usually are not caught off guard when traffic volumes increase. It is why they add capacity to begin with. 

Past forecasts of vehicle miles traveled in the U.S. compared to actual trends. Source: Frontier Group

Many of these forecasts rely on projecting past trends into the future. Because the historical rise in vehicle use was fueled by massive highway construction and outward growth, these forecasts also assume the same patterns of infrastructure investment and urban development. They perpetuate induced travel from the past into the present.

Most travel demand models are not designed to break this cycle. Land use policies and future development patterns are typically decided at the municipal level, sometimes in coordination with metropolitan planning organizations. Those growth patterns then feed into travel models, compelling transportation agencies to meet the anticipated demand—i.e., widening highways that serve suburban growth on the urban fringe.

The anticipated development comes to fruition, and the cycle continues.

Some professionals prefer to call this “latent demand,” which implies that major highway investments unleash unmet economic growth and social mobility. Research by Reid Ewing suggests that is not the case. An important question, therefore, is whether unbridled growth in vehicle travel is a good thing, especially when it translates to poor health outcomes, more pollution, worse traffic congestion, and higher transportation costs.

When asked how to tackle this issue, experts at TRB focused on several types of improvements, summarized below.

  1. More advanced models. Traditional models (often called “four-step” models) are plagued by shortcomings. Advancements like activity-based modeling, dynamic traffic assignment, and—perhaps most importantly—feedback loops for land use interactions can make them more responsive and more accurate. But it is important to remember that any model is only as good as the underlying inputs and assumptions.
  2. More transparent modeling. Just like weather forecasts, no traffic forecast is perfect. So, when major infrastructure investments are involved, it benefits transportation agencies to be as transparent as possible in how they are predicting the future. That could mean providing open-source models, clearly documenting model assumptions, or providing information about model accuracy and margins of error. Agencies can also archive past forecasts for easy public access, and regularly assess how those models performed.
  3. Focusing on tangible outcomes. When it comes to choosing and designing projects, well-calibrated models are just one tool in the toolkit. While most models are geared toward the future—often decades out—there are plenty of projects that could benefit people today by improving safety, increasing access, and supporting local economies. Accessibility analysis is another promising tool already being used in Virginia (alongside travel demand models) to prioritize investments that connect people to jobs and other opportunities by all modes.

While data and models are critical for informing investment decisions, it is important to note that those decisions are ultimately value-driven and should reflect each community’s vision for the future. Just as highway projects induce more driving and outward growth, other types of investments can “induce” transit use, transit-oriented development, safer walking and biking, or any number of associated benefits. Achieving those outcomes means not just improving model accuracy, but also choosing the right tool for the job.

Photo Credit: Sajjad Ahmadi via Unsplash, unmodified. License.